I joined the Google Plus development team shortly after the project began in 2010, moving from the Blogger team. I participated in the project for three years, first working in the profile development team, then moving to the Growth and Engagement Team (GET), and finally ended up in Project Madonna and Project Zorro, where I helped to end the “real name” policy .



I was a grassroots fifth-level software developer, so I took a rather insignificant part in the project, but witnessed many decisions made during the initial development phase of Google Plus, and it seems to me that I can point out critical errors made in the early stages of the history of this social network. It’s not only that we are all strong in hindsight - even then I had a clear feeling that in the long run these decisions would be harmful. Since then, only the fact that I can now present my arguments more intelligibly has changed.



I won’t argue that the problems I described were the only ones that Google Plus encountered, but I think they made a significant contribution to the gradual fading of the project. All of the following is my personal opinion, not the current employer or other persons.

Asymmetric Subscription Model

A symmetric subscription model (according to the terminology we used at that time) is a model in which both parties agree to establish a relationship between each other. This model is used on Facebook and LinkedIn.



In an asymmetric subscription model, one of the parties single-handedly establishes a connection by subscribing to the other without its participation. This model is used on Twitter and Google Plus.



The founders of Google Plus sought to rapidly grow the graph of social connections. They knew that defeating Facebook was a matter of scale. Ultimately, people come not to those social networks where functions are better, but to those where all their friends are already there.



They believed that the asymmetric model would provide a faster growth in the graph of social ties, because it does not require an “agreement” to be friends on both sides. And they were right - at least at first.



At the same time, they wanted Google Plus to be a network of “close ties” that users would use to communicate with friends and family.



However, the asymmetric model has a key weakness - it is a poor channel for two-way communication. When you “subscribe” to someone, it serves as a signal that you want to listen to what this person wants to share. But if you subscribe to someone, this does not mean that he will listen to you.



Such a unidirectional model is actually a bad choice for a close network because close relationships depend on frequent two-way communication.



"Hi, little sister, how are you?"



"Not bad, how are the children?"



I can add my sister Christie to my Google Plus circles, that is, I’ll start to see the content she publishes. However, there is no guarantee that she will see what I publish.



People are motivated to communicate. But they are motivated to do this only when the other person is actually listening to them. Screams into the void extremely demotivate most people. What else demotivates is the lack of feedback. If there is no signal informing that the interlocutor has received your message, then the natural human reaction will be to stop using such a communication channel and search for a new one.



Already in the early stages, it became clear that the messaging in Google Plus was very unreliable. Not in the sense that Google Plus was losing the message or couldn’t deliver it, but in the fact that there was a high probability that no one would waste time reading it.



There is no such problem in email - although you don’t see a signal that the recipient has read your message, you know how the mail client works; you know that the message will still remain in his “Inbox” until he does something, and there is great confidence that the message will be read sooner or later.



This is not a problem on Twitter, because it is not a network of close ties. On Twitter, communication is a statistical phenomenon: you don’t care if the message will be received by each of the subscribers; only the number of subscribers and the fact that a large percentage of them received a message is important to you.

Rank Based Mapping

Another factor that contributed to the overall insecurity of communication in Google Plus is the use of ranking. When a user reads content in the Google Plus stream, this content is ranked by a complex algorithm, so that “important” or “interesting” content appears at the top and communication with a lower priority is displayed on the page below.



In addition, the page supported “endless scrolling,” meaning it had no end. Continuing to scroll through the page, the user opened more and more posts. This meant that he could never “finish”. There was no finish line, a sense of completion. This in turn meant that each individual user had to stop sooner or later, and the place of this stop was unpredictable.



The combination of these two factors was, in my opinion, a killer blow for the implementation of "close ties." If my sister sent me a message, then there is no guarantee that it will be at the top of my stream, and if it will be lower, there is no guarantee that I will get to this place before I finish reading. And this is no longer a problem for me, but for a sister who will not know if I read her message.



The ranking problem was that it took away the reader’s ability to control. In the end, who should decide what is important and interesting and what is not? The Google Plus project leaders were convinced that ranking was the solution — ultimately, all of Google’s success was based on ranking algorithms — and that any problem could be solved by coming up with a better, more personalized algorithm.



Ranking can be very helpful when viewing information with a low priority (for example, when reading a newspaper), because no one wants to make hundreds of decisions about what to read and what not to read. In this context, the existence of an automatic “curator” is justified.



But in the case of important things, such as letters from mom, the user wants to have full control over what he reads. (That's why Google Inbox never took off.)



And users quickly realized that when sending a message to a friend or family member, there is a high probability that he will never pay attention to him. Therefore, they returned to email or Gchat.



Subsequently, the company added a “notification” function to Google Plus, which had a separate user interface that is visible in all Google applications. It displayed a list of messages designed specifically for the user or, for some other reason, had a high priority.



But this function had two problems: firstly, it simply appeared too late; secondly, by creating a separate independent interface instead of eliminating a fundamental problem, Google weakened the UX as a whole. This meant that if you want to use all the features of Google Plus (both stream and private messages), then you have to read many messages twice.



It is also worth mentioning some other problems of ranked endless scrolling. Since there was no “mark as read" button in the interface, the only way to make sure that you already read the post was to track whether you scrolled it. It is clear that this is a very unreliable signal. In the case of a false positive, the user thinks that he has already read the article and will no longer open it; in case of false negative, the same post will appear again and again.



This happened because the user interface designers wanted to simplify the interface as much as possible and not force the user to click Finish on each post. It seems that they were so fascinated by their own ingenuity that they did not see the shortcomings of such a solution.



Of course, in the Facebook news feed, many of the design elements described by me are used, including ranking. However, important messages sent by one person to another have a separate channel, more similar to e-mail: it has a strict chronological order, rather than ranking, and the explicit function of "mark as read". Ranking and algorithms apply only to not so important posts.



Facebook designers clearly understood that when it comes to social interactions, there is no single universal environment suitable for all situations. There are differences between essential and non-essential communication, between urgent and everyday, especially when it comes to maintaining interpersonal relationships, and they can not simply be thrown into the general list for the sake of simplifying the user interface.

Inorganic growth

Despite the problems described above, immediately after the launch of Google Plus, this network actually seemed pretty fresh and vibrant. There were many enthusiastic users and diverse points of view on interesting topics. I remember my admiration when I first had the opportunity to partake of the wisdom of Will Wheaton. The network was growing - slowly, organically, but still growing.



However, everything changed when the company began to drive Google users into Google Plus in droves.



Again, management was concerned about scaling in order to defeat Facebook. And she already had billions of users, that is, people with Google accounts. How do I get all of these Google users to start using Google Plus?



To solve this problem, a project was created under the internal code name Project Hancock. He had to create a Google Plus account for every Google user. In fact, everything was much more complicated than it seems; To implement the project, it took about three months of the development department’s work.



All of a sudden, millions of new users appeared on Google Plus. And many of us have begun adding these users to our Google Plus circles. But there was a problem - many of these new users did not respond and in no way interacted with other users.



Soon after, I began to notice that people began to call Google Plus a "ghost town." Only the network was not a ghost town in the traditional sense. Rather, it was a city inhabited by ghosts. More specifically, ghost accounts.



As you can see, Hancock was supposed to create millions of new Google Plus users, but instead created millions of accounts. And these accounts have become a problem.



As part of my work at GET, we collaborated with a nearby analytics department. Each week, we watched a one-hour presentation prepared by one of the analysts with a deep analysis of some statistical metric or analysis. It has always been interesting.



One of the presentations was devoted to the effect of inactive accounts. Logically, we assumed that “more friends = better”. That is, the larger your circles, the higher the likelihood that you will become an enthusiastic user. However, the data showed a more controversial picture. In fact, it turned out that "more active friends = better." In fact, there was a slight negative correlation between user engagement and the number of inactive friends in the user's circles.



In essence, this meant that if you have a lot of inactive friends in your circles, then this creates a bad impression, leading to the abandonment of the use of the network. In fact, it is better that the circles are small and limited only to active friends. Unfortunately, users did not have an easy way to figure out which of their friends is active and which is not.



In a symmetric subscription model, such as Facebook, this would create fewer problems, because in order to “freeze” a person, he must be active and confirm the offer of friendship.



In my opinion, Project Hancock was the beginning of the fall of Google Plus, after which the network could not recover.



The combination of these three factors - asymmetric subscription, excessive confidence in ranking and a large number of inactive accounts - inspired users a sense of uncertainty when writing messages to family and friends. On Facebook, I can start a quick personal conversation with my sister, and we both will have great confidence that our messages will be received and read.



In the case of the Google social network, I was forced to leave Google Plus and use another application, such as Gmail or Hangouts. More importantly, I had to step out of the circles / subscription model and use a different set of connections, such as Gmail contacts. This flaw destroyed the possibility of creating a social network of "close ties."

Meta: group thinking

Google has some of the smartest people in the world. Why didn’t they notice these problems from the very beginning and solve them?



In fact, many saw them and tried to convince the leadership to change course, but to no avail.



Around the beginning of 2011, I had lunch at one of the Google cafes with Vic Gundotra, the head of Google Plus. In a normal situation, Gundotra would not hang out with such a mid-level developer, but he found out about my previous career in game development and was curious.



He directly asked me what I would change in Google Plus, and I said that the model is asymmetric subscription. He indignantly rejected the idea, arguing that, in fact, the asymmetric subscription model was one of the best decisions made in the project.



I mentioned other problems - ranking, endless scrolling and inorganic growth, which other employees also complained to their managers and various responsible persons. This is not to say that our objections were rejected without hesitation: the people responsible for the vision of the product had detailed logical arguments in favor of the fact that their method of implementation was the best.



At the same time, we could not be completely sure that we were right. We could not provide concrete evidence that the chosen course was erroneous. And when the leadership is extremely smart people who earn ten times more than you, then you usually give them a loan of trust. After all, they must definitely know what they are doing.



But these leaders were deeply committed, both intellectually and emotionally, to their own specific vision of the product. And they were surrounded by equals to them, who also believed that everything was being done right. As a result, we got a classic group discussion situation. There was a strong tendency to do everything in a concrete way, for example, to rely on smart and complex algorithms in matters of choosing the best content for users instead of giving them the opportunity to choose for themselves. Indeed, such an approach has always worked for Google and, to a lesser extent, for Facebook.



Over the years, it actually became increasingly clear that some of the initial design assumptions were erroneous. But then it was too late to change anything. Recycling the subscription model after millions of links have already been created would be damaging to users. Following this path, we could not go back.

Effects

By 2014, it was clear that Google Plus would never be the “new Facebook”. I took part in one of the many open discussions on strategies being led by new project leader David Besbris. Unlike his predecessors, he seemed open to the perception of the idea of ​​fundamental product change.



However, their chosen strategy was rather passive. In fact, Google Plus was going to completely abandon the goal of becoming a network of "close ties." Instead, management decided to redouble its efforts to turn it into a network of “close interests”, that is, communities of people who have common passions but do not necessarily know each other in real life. This was based on the fact that many of the interest communities in Google Plus were still quite active, and therefore, the product could potentially become successful by focusing on this audience.



In other words, instead of striving to become the new Facebook, they aimed to become the new Tumblr.



By this time, I had already left the Google Plus team and worked in the group of JavaScript systems; although most of my work was related to Google Plus, I was not on the product development team. I left Google at the end of 2015, so I don’t know what happened after that, but according to the information reaching me, the social network gradually faded away until it was finally closed on April 2 of this year.



I wanted Google Plus to succeed. As many have noticed, it had a lot of attractive features, including many subtle aspects of design, which turned out to be better than Facebook and any other social network. And I sincerely do not like Facebook, both as a product and as a company. My reluctance to use Facebook means I have a lot to miss.



I think many people are just as unhappy with Facebook as I am. However, the idea that someone will come and create a new social network that does not have the disadvantages of Facebook seems hopeless. In the end, if even Google with its huge fortune and talented staff did not succeed, then who else can do it?



But thinking so wrong. Google Plus failed not because Facebook is invulnerable. It failed because of the fundamental flaws inherent in it from the very beginning. And learning from these mistakes will be the first step towards creating something better.